Advanced Imaging Medical System

Powered by computer vision, ML, and powerful LLMs

NOTE: WE HAVE RE-BRANDED TO AIMS HOWEVER PRE-RECORDED CONTENT ON THIS PAGE MAY STILL SAY "ASTROMED" OR "MEDICINE-AI" OR "MODIFIED MEDICAL"

Problem Statement

“Astronauts need a reliable, time-efficient, and non-obstructive way to determine exactly the medical items they have onboard, the location of said items, and the quantity of said items. Current ISS solutions are for the most part manual, not time efficient and are not fully integrated into all the medical operations of the ISS. With this project, our task is to design a modified medical inventory system that hopes to improve accuracy, speed and autonomy while meeting all the constraints of a spacecraft.”

Requirements and Constraints

Requirements:
Must track quantity of medical items
Track location of medical items
Provide alerts for missing/low stock of
medical supplies
Log item retrievals correctly
Must be Earth-independent
Have user-friendly UI for astronaut use
Be able to endure heavy density of data with scans
Needs to be a lot more time-efficient
Incorporate a sufficient authentication
system w/ Facial Recognition

Constraints:
Cannot interfere with ISS operations
Needs to fit inside medical lockers
Needs to work with repackaged
medications
Medications hold different quantities (ie;
60 pills in a bottle vs. 1 EpiPen)
Must authenticate astronauts
Must practice proper medical privacy
guidelines

Research Results

Space is an unstable environment susceptible to rapid injury; and in such cases, organization and time
efficiency can be the matter of life and death
The current system employs partial RFID implementation, but this system is not granular or time-efficient
To try and aid with this problem, NASA developed the Medical Consumables Tracking System (MCT)
The MCT System only scanned one pack at a time, every 720 hours, and had no authentication (facial/speech)
To resolve this issue, we started to develop a modified Automated Medical Inventory System

Intro to AstroMED

Our Introduction video

Our Solution

A Computer Vision with AruCO codes (markers that can be scanned to track quantity and medication) is implemented due it its ease of installation, modularity (simply a sticker) and ease of implementation in terms of hardware, since all you need is a camera.
A webapp on a tablet on the ISS will be used on an interface by astronauts, this is where the speech recognition and user interface will live and allow for the system to be Earth-independent
Through facial & voice recognition, we will be able to know who takes out what and tracks medication
MedicineAI is incorporated to make accurate predictions about medication expiration dates and resupply dates.
Advanced Inventory Control with MedicineAI
STTS voice recognition module brained by powerful LLM (large language model)
The ground control will be able to connect using another redundant webapp incase things go wrong.
For our hardware, we use a custom shelf and box design that allows for efficient locking of the boxes using only one solenoid per shelf unit.

Recent model redesign

It is important to note that we recently redid our CAD models to better represent the new pill recognition system. Attached below is an isometric projection of the new models